AD-All? 108 ARMY ELECTRONICS TECHNOLOGY AND DEVICES LAB FORT MON-ETC F/9 10/3HIGH1 ENERGY SULFURYL CHLORIDE BATTERIES. 1

JUN 82 S GILMAN, W WADE. M BINDER

UNCLASSIFIED E l

aW

jut, 196

GILMAN, WADE, .BINDER

__ HIGH ENERGY SULFURYL CHLORIDE BATTERIES

SOL GILMAN, WILLIAM WADE, JR., aad MICHAEL BINDERUS ARMY ELECTRONICS TECHNOLOGY AND DEVICES LABORATORY, ERADCOM

FORT MONMOUTH, NEW JERSE7 07703

"_ _ . INTRODUCTION-_

here is no practical alternative to the use of primary batteries topower man-portable electronic equipment for communications, surveillance,target acquisition, and Night Vision applications. For an increasing per-centage of such newly-developed equipments, rower and energy densityrequirements are so high that only the most energetic electrochemicalcouples can be considered for the purpose. Cells utilizing sulfurylchloride as the cathode reactant are the latest and most energetic of the"liquid cathode cells" resulting from research initiated at the Electronics.Technology and eevices Laboratory in the early 1970s.4--

Table I lists "full-cell" reactions and other electrochemical inform-ation for 5 different primary cell systems.

TABLE I - ELECTROCNMICAL CELL REACTIONS

Theoretical TheoreticalPotential Experimental Energy Density

SYSTEM CELL REACTION E V _ OCVy) , Ib

M/NO 2 MR + MO 2 + H20 -. Mn2 0 3 + Mg(Off) 2 2.8.l 2.0 324

Li/SO2 21.1 4 2SO2 - L 2 S204 ---- 2.95 611'

LI/SOC12 41 + 2SOC12I--a- 4L.1 + S + SO2 - 3.65 668*

L1/SO2C1 2 21.1 + So 2C12 -- w-2LIC1 + So2 3.909 3.91 639

3:1 CaIs/ 2C12 Ca + so C1,- CaCI 2 * S2 3.813 3.30 584

*Based on open-circuit voltage.

The first system is the Army's present aqueous magnesium cell whichfollows conventional electrochemical practice in that the anode and cathode,reactants are both solids .and mechanically_ separated.to-.interact only ....

D T C DISTRIBUTION STATEMENT AB &plovqd t publt rlea;%

SI IELECTE D Uj=e ... L,

- 82 07**JUL21 I

, I , , , . ... ,-,.+. .

GILMAN. WADE, BINDER

through an external electrical conductor. The other table entries are---for "liquid cathode" cells which successfully violate the rule of sepa-ration of reactants. Although the anode and the cathode reactants are indirect contact (as the latter are liquids), rheir direct chemical reactionJs kinetically hindered by a thin and protective layer of salt on theanode.

The relatively low theoretical energy density of the aqueous mag-nesium cell trcnslates into a practical energy density of approximately45 Wh/lb at moderate rates of discharge. In the nonaqueous systems it ispossible to utilize much more reactive anode:i and cathodes than in theaqueous magnesium system; therefore, higher electrode potentials, experi-mental voltages, and energy densities are feasible. For instance, theLi/SO2 cell field-tested by our Laboratory aLtually provides an energydensity of 100 %h/lb at moderate rates of di:charge. The Li/SOCl2 cell is" our first "oxychloride" system; it is already undergoing 6.3 stage develop-

- ent for target acquisition appl1icat ions. .- can --el iver -appro ximat .el-y"150 Wh/ib at moderate discharge rates. The ;.i/SO 2C12 cell system, which

will be discussed more extensively in this paper, has a theoretical energydensity slightly lower than that of the Li/SOCI 2 cell. However, the higherterminal voltage and larger practical cathode capacities, which we havebeen able to obtain, allow us to project denrities greater than 200 Wh/lbfor a practical cell.

Several years ago, it was taken for granted that energy densitiesgreater than 100 Wh/lb could be obtained only with cells utilizing lithiumanodes. The lnst entry on Table I reveals that a Ca/S0 2C12 cell systemhas the potentiality for performance in the same range as that of thelithium cells. Encouraging preliminary results on that system will alsobe reported here.

EXPERIMENTAL PROCEDUB ES

iPreparation of Cells: Preparation of the electrolyte, electrodes, and,assembly of the cell is discussed elsewhere (1). All steps and procedures!involving the exposure of either the clean anodes or of the electrolytewere performed either in a glove box (argon atmosphere) or in a dry room.The preliminary "screening" of several different carbon black powders wasperformed on "uacompressed" electrodes made by a "standard" (1,2) tech-nique, while the optimized United carbon electrodes were made using acold-compression step in the fabrication process (1).

The cells were assembled using one 2 cm X 2.5 cm cathode sandwichedbetween two anodes of the same size with fiberglass filter papers provid-ing mechanical saturation between the electrodes. A small lithium elec- C1

trode was incorporated into the cell to serve as a reference electrode.All current densities are based on the total current divided by 5 cm2.The cells were horizontally oriented on the bottom of an open teflon con-"taner with 5 cm3 of electrolyte. The teflon container was, in turn,placed in a gas-tight glass outer container with electric.leed-throughs......-

odes

.)i~'L Speclal

• ,+:+ .z --...- I

I. i L :

GILMAN, WADE, BINDER

Electrochemica] Measurements: Discharge and polarization-curves were

measured at 22' ±2°C using a constant current power supply. The polar-

ization curves were measured by applying pre-determined constant currentsstarting from 0.04 mA/cm2 and allowing 3 minutes before recording each

closed circuit potential. All potentials were measured against the lithium"reference" electrode.

Cathode Porosity Determinations: The percent.porosity is defined as the

percent of wet cathode volume available for absorption of SO2CI. The

volume of S02Ci2 was determined by weighing a cathode before ang after

immersion in SO2 CI2 and "blotting" on a glass surface. The wet volume of

the cathode waf determined by measuring its linear dimensions with

calipers.

Determination )f "True" Surface Areas: The surface area of carbon black

powders and of complete electrodes was measured by the Brunauer, Emmett, - 1

Teller (BET) method using an Orr Model 2100 Surface Analyzer.

Electrolyte Corductivities: Conductivities were measured using an AC

bridge and a thermostatted glass-stoppered cell with a cell constant of0.200 cm-1 .

RESULTS AND DISCUSSION

Sulfuryl Chloride Electrolyte Solution: Of the salts with known solu-

bility in SO C1 , LiAlCl has long-term chemical stability in solution,has a solubility upwards of 2 moles/liter and does not undergo noticeabledirect chemical reaction with lithium. Conductivities of a 1.5 molarsolution of LiAIC14 in S02 C12 are plotted for a wide temperature range in

Figure 1.No precipitation of the solute o. 5

.was noted down to -54C (the freez- _-

King period of pure SO Cl is -54.1C). E ."At room temperature, he specific -1

conjuctivity of approximately 0.OlJLcm is of the same order as that 00"

encountered for the electrolyte used PoAwith other high energy battery sy- go.

stems. When using such electrolytes, a..

internal voltage drops in cells arekept small by restricting separator to -0 tO 0 t 20

and cathode thickness to less than a

millimeter. The low conductivities Figure 1. Conductivities of 1.5

encountered at -20*C or lower do M LiAlCl4 - S02 C12.affect performance significantly andimproved electrolytes are being sought in our continuing in-house program.

p:!

GILMAN, WADE, BINDER .

Lithium Anode iii Sulfuryl Chloride Cells: A lithium-electrode with a ___1

freshly prepared (by mechanical abrasion) surface shows little voltage* - polarization for anodic current densities up to 40 mA/cm. On current

reversal, dissolved Li+ is reduced to metallzc lithium and deposited on, the surface of the lithium electrode in a highly dendritic form. Although,

:precise thermodynamic measurements have not been performed, one may con-'clude that a fresh lithium electrode is closeto reversible in a solution

of LiAICI4 in S02C12 . The stability of the lithium electrode in SO C12 is,due to the formation of a self-limiting thin coating of LiCi, formed spofl-:

taneously according to the reaction in Table I. On long-term storage, that• . . film apparently thickens and would introduce-a" -voltage delay" problem in

. . battery cell. This potential storage problen is presently under investi-

gation by a contractor (3).

1i;n, Teflon-Bonded Cathode in the Li/S02C12 Cell: _Electrodes for lithium- thionyl chloride cells are normally formulated with Shawinigan carbon black

and Teflon emulsion (normally 5-10% TFE in t'le dry electrode). Prior:results for sulfuryl chloride cells, using similar electrodes, produced A

rather discouraging results (4). In the present work, we experimented withseveral different types of carbon black powders and with the process ofelectrode fabrication. Table II lists the carbons studied along with,properties of the original powders and of completed electrodes fabricatedusing 16% TFE and a standard fabrication technique.

TABLE 11. Teflon-bonded Carbon Cathodas (16% TE)

Carbon Per grau Per cm2 of

loadine of carbon electrode geo- 2 electroec

of Carbon Derivation (/CU,, powder metric area* porosit*_

Sbw Liagat--502 compresed Decompoeition of acetylene 0.0194 66 1.28 87

Darco-G60 fteam-activation of charcoal 0.0048 301 14.4 64

United XC-6310-4 Decoaposition of oil 0.014, 1000 14.6 81

Columbia HK 1670 Decmposition of oll 0.0104 1200 12.5 ?5

a BaiAd an (length x width) area of electrode, one side.

The resulting electrodes have "true" (BET) areas spanning more thanan order of magnitude and electrode porosities ranging from 64 to 87%.

Figure 2 presents polarization curves for such electrodes, with currentdensities based on the "superficial" area of the electrodes. From the

" 2 figure, it is clear that very significant differences exist between the

V . . .I :

(" 7

GILMAN, WADE, BINDER

electrodes, with "United" carbon black affording the best initial voltage -

over most of the current range. The results are replotted in Figure 3after basing the current density on the "true" (BET) rather than on the

"superficial" surface areas. On that basis, it may be concluded thatalmost identical results are obtained for all. of the carbons in the lowerrange of current densities where mass transport effects may be expected

not to play a dominant role. This is consistent with the surface-chemical'

mechanistic sequence to be discussed below.

AA

~N A United XC-6310-4 \.'5 V Shawnigan-507 Compressed

o D ̂ACO-o-60 2 0u O Columbia ER 1670 \ 0

Current Density, m /m 2 (BASED ON GEOMETRIC AREA) Current Density. mA/cm

2 (BASED ON B..T. AREA)

Figure 2. Polrization curves for Tef- Figure 3. Polarization curves forIon-bonded carbon cathodes in lithium- Tefton-bonded carbon cathodes insulfuryl chloride cells. Current den- lithium-sulfuryl chloride cells.sities based oil "superficial" cathode Current densities based on "true",areas. (BE surface areas.""

Figure 4 shows that the capacity of the test cathodes depends directl Iion the gross electrode porosity as measured by solvent absorption. The

,porosities are determined, to some extent, by the inherent microscopic;"structure" of the carbon powders. Shawinigan black, for example, is noted.for its highly developed acetylenic structure, carried over from the mo e-

cule from which it is originally derived.SFrom Figures 1-4 one can conclude that the "best" cathode would com-

':bine both high BET area and high porosity. Attempts were made to accom-plish these conditions, simultaneously, by formulating cathodes with mix-tures of Shawinigan black and the higher area carbons with unsatisfactoryresults. Success was achieved by varying the fabrication process in orderlto increase the porosity of United carbon eldctrodes. Figure 5 presentsrepresentative discharge curves for different formulations. The use ofUnited Carbon and a "cold-compression" process (Trace 5, Figure 5) wasadopted as standard for the balance of this work. The electrode wasfurther optimized with respect to teflon content (at 11% TFE). A t picalelectrose lectrosity of 87% as carbon loading of 0.024 gm/cmTh

" . ..por.sitie..are.determined,.to.some..extent,.by.the.inherent.microscopic"srutre f h cronpodrs haiignblck oreaml,"snoefor its highly developed acetylen~~i c tutrcridoe rmteml-

cue rm hchi i oiinlyeivd

From.Fiures.1- on ca"ocue htte"es"ctoe ol on

GILMAN, WAD, BINDER .

A/~~ --- --

W 40 -"f_ ..~~0/_ ,/..i .. ..........................- ...

S o DARCO-G-6020 OCOLUMBIA H 1670 .

A UNITED X-6310-43: ./. *V S.KAWINIGANr-50 COMPRESSED I U'I!ED (AGO. "LAIL, (O'PISS|

50 6 J 80 90 4.. 2 •

PEPCEN1 POROSITY .... M., K

Figure 4. Capacity-porosity rela- Figure 5. Discharge curves at currenttionship for T,:flon-bonded carbon drains of 5 mA/cm2 for differently-

,cathodes (standard, "uncompressed'; formulated cathodes in lithium-sul-fabrication). furyl chloride cells.

Figures 6 and 7 compare performance of the cathodes of our Li/S02CI2

cell with that of the older Li/SOCd2 system, which was previously the mos':energetic ambiEnt temperature primary cell system know.I. Because anodepolarizations are small, the voltages are essentially those of a completepractical cell. Although the performance illustrated in Figure 6 and 7 if;excellent, there is still the possibility for significant improvements athigh currents end low temperatures and that is one thrust of our continu-ing battery re.earch program.

a -"

9. a (2) C-0.O1, .- 2C , I - .6/.'€

() C.o l,h. .ch , -SO.2.I - 20 fV€,, 2 1

SRI, 1.I l 1 .€ 2 6 6 IQ Il 1, 16 Is 2) d2Cn.r n, . 6. .? (.a2 wm. . WRRtI M.l) I . (_'),

Figure 6. Comparison of cathodic Figure 7. Comparison of cathodic dis-polarization curves for lithium- charge curves for lithium-sulfurylsulfuryl chloride and lithium- chloride and lithium-thionyl chloridethionyl chloride cells. cells.

. . . . . . . .

vI

GILMAN, WADE, BINDER

Electrode Dischirge Products in the Li/SOC 2 Cell: The overall dischargereaction listed for the Li/S02Cl2 cell in Table-lwas deduced throughx-ray diffraction, gravimetric chloride determinations and volumetricdeterminations of SO2 gas release. First, it- was determined that LiCl is,practically speaking, insoluble in the electrolyte solution employed here.Hence, discharged cathodes could be exhaustively washed with pure S02C12

Ll to eliminate entrained LiAlCl4, without solutbilizingLidl formed as aproduct in the cathode. Diffraction patterns for cathodes which werecarefully washed, dried and crushed revealed only cubic LiCI. Two cellsmade wi~h United carbon cathodes were discharged at a current density of5 mA/cm to a citoff voltage of 2 Volts. The cathodes were washed with

t*. SO2C12 , dried and quantitatively extracted with water. Aliquots of theaqueous solutions were potentiometrically titrated for CI- using a stan-,dardized AgNO3 solition. A small correction for occluded AICI was madeby determining Al+._colorimetrically and subtracting four times_that num-l.iber of equivalents. The corrected equivalents of Cl were found equivalen",to the coulombs of electricity passed, to wit in a 2% tolerance. Thisiestablishes that LiCl is the only ionic product of cell discharge and alsothat the LiCl is quantitatively precipitated in the pores of the cathode.

To evaluate non-ionic discharge products, it was first determinedquantitatively that gaseous S02 is released during cell discharge. A cellwas then assembled with electrolyte which was pre-saturated with S02. Thecell was attached to a gas buret (using fluorocarbon oil as the displace-ment liquid), and discharged at 5 mA/cm2 whij.e monitoring the volume ofgas released. in duplicate experiments, it was established that one moleof gas was produced for every two equivalents of electricity passed.

The overall reaction listed in Table I is the simplest one consistentwith the analytical results reported above. The following individualelectrode reactions are consistent with the overall reaction:

anode: 2[Li-Li ++ e-] (la)

cathode: 2Li+ + SO2C12 + 2e-- o2LiCl + SO2 (lb)

overall: 2Li + SO2C12 -- 2LiCl + SO2 (2)

E* - 3.90) V (30c)

As already discussed above, the capacity of the cathode is proportional toits porosity. As is apparent from these findings, the porosity is requiredin order to accommodate the LiCl precipitate while allowing continued goodtransport of the cathode reactant and of conducting ions.

The analytical results reported above contradict earlier reports (5)that Li2SO4 and molecular sulfur are major products of cell discharge. Itis felt that the earlier results may have resulted from a failure to avoidwater contamination early in the analytical procedure. The present resultsare particularly significant for their relevance to battery safety, inasmuch

u . .I,', = . , -

GILMAN, WADE, BINDER,

as molten sulfur can react vigorously with lithium under conditions of ;-cell malfunction. This is potentially a very significant advantage of theLi/SO2Cl2 over the older Li/SOCI2 system (see Table I).

C Cathode Mechanism in the Li/S02 C12 Cell: Suifuryl chloride is known (6)

to dissociate into S02C12 according to the fvllowing reaction:

----- s. 02 C12S 2 +C1j- (3)1

Both gases are appreciably soluble in our el(.ctrolyte (1.09 and 0.62 molalfor SO2 and Cl2 , respectively, at 24*C as determined in this laboratory).Furthermore, as determined by.W. Behl (7), of this laboratory, molecularC1 is more reactive than undissociated.SO2C'2 . Figure 8 shows that satur-ation of the eloctrolyte with either SO2 or Cl2 produces a noticeable shiftin the cathode )olarization curve. The particularly marked effect of SO Iimplies that the latter.is-acting by.removing.Cl2.as..cathqde..eactantani!shifting the eq.iilbrium-of-reactiom.(3) tot.he left. A complete cathode,mechanism, consistent with our observations, would involve the productionof Cl2 by reaction (3) followed by the cathodic reduction of C12 :

C12 + 2e- + 2Li+- 2LiCl (4)

Carbon is a known catalyst for reaction (3) and that would explain the im-portance of higi electrode surface area to good electrode performance.

Quantitative information on reaction(3), in the condensed phase, is de-

TEL CR,,, T ::TT" 'U. sired for confirmation of the pro-£C c&hO,, IA,,a..ITHSO posed cathode mechanism and as a

basis for future cathode improve-9 P_ T -ments. Preliminary studies have

.. ..-- *-*--.-. ". been performed using ultraviolet38- spectrophotometry. Figure 9 shows

'4.,'u that "aged" sulfuryl chloride ab-

. . . sorbs light with wavelengths less*0.1 ,oo than 460mu(the liquid appears yel.

CUINT 0108171 KVJAi (BASCO low). The absorbance is diminished

Figure 8. Cathodic polarization by freshly-saturating with argon, Icurves of lithitm-sulfuryl chloride and further diminished by saturatingcells containing gas-saturated elec- with SO2 (the latter does not absobtrolyte. light in this region of the spec-

trum). The absorbance is intensi-fied by adding C12. Comparison of the results suggests that the "aged" so-lution has undergone decomposition. Estimates of the rate of decompositioncan be made by recording the absorbance (e.g., at 400 m) vs time relation.ship for freshly degassed samples. Preliminary results reveal that, at30*C, the homogeneous rate for the "neat" solvent is equivalent to lessthan 0.2 mA/cm! (exact rates depend on solvent purity) and, hence, wouldnot support good cathode performance at high current densities. In the

II

!, q'.

GILMAN, WADE, BINDER

presence of carbon, however, the estimated htterogeneous rate of decompo-

sition is sufficiently large to account for the full range of current den-

sities covered in Figure 8.

a., nsI.

U \ .. .

.' - -Figure 9.--Ultraviolet spectra of .-sulfuryl chloride samples.

-Performance of Electrodes in the Ca/SO CI Cll: The overall discharge ireaction listed for this cell in Table-as based on analogy to theLi/SO C12 cell. For that assumed reaction, the thermodynamic cell poten-tial is only 91 millivolts less than that foi the analogous lithium system.From Figure 10, it can be seen that experimental cell voltages are alwaysat least Vole less than the thermodynamic value of 3.818 Volts. Further--more, most of ':he low current density polarization occurs at the calciumanode, which m-L'ght otherwise be expected to assume a potential approximatc-ly 90 millivolts positive to the lithium reference electrode. However, itis highly encotraging that the calcium anode shows little tendency to.undergo additional polarization at high current densities (i.e., it holdspromise for good "high-rate performance). The cell polarization at highcurrent densities is almost entirely attributable to the cathode, similar,to the situation for the Li/SO2C12 cell.

40 CAT0O1( VS LIThIUM RIFI US

..............

as

' Figure 10. Polarization curves for.. , -. a calcium-sulfuryl chloride cell.ILO

CANCTY mA-1

to IS Lf!! =

Figur 10. olarzatio curvsfoA. t a clciu-sulurylchloiderell

AFIMY SC IENuCE CONFERENCE lNAPER

-- STAMP Securityn ,',it list Classification

nan.' ol au- herethor(s) he, c GILMANKADE-- GI MAN ,___-P LBI-N-DER

Start herefor all

page-s -- ___.Figure 11 compares cell polarizations for our Ca/S02Cl2 cell withaftcc the those for our Li/SOaCl 2 cell and with a commercial Li/SO 2 cell of the type

first presently being used in Army radio sets. From the figure, it is clear

that the new calcium cell, while still inferior to our best lithium tech-On first nology, bas the potentiality for displacing the Army's present lithium-

pae ty,1 , sulfur dioxide cell for at least some atplications. This moy opply parti-

tit Io Of cularly where the greater chemical stability of calcium out%'2ighs thepaper here superior electrochemical performance of lithium.

A u h r *4.00 ................... ........ *........................................ U / so C1Author -- .. .. .. .. .. .. .. ..... o tAffil iatiou !City, State 3JO-

".-.. ..... ...................... c /Sq.%:First line .........."

1.60- L

120-

0.1 to I00 100.0

CURRENT D0'NSITY, mA/ma

Figure 11. Polarization curves for three dif-

ferent high energy cells.

Although the polarization characteristics of the electrodes in a

calcium-sulfuryl chloride cell are highly encouraging, we are prescntly

faced with the serious problem of reduced cathode capacity (i.e., life) in

any electrolyte containing dissolved calcium ions. Presently, cathode life

is one-half or less of that for the analogous lithium cell. Micrographs of!

spent cathodes from the two types of cells (Figure 12) provide a rlue to

the origin of the problem. In the lithium cell, the LiCl which forms in

the cathode during discharge, is deposited as multiple discrete clusters ofl

cubic crystals which allow relatively easy passage of ions, reactants and I

products from the liquid electrolyte phase. By contrast, the solid product,

phase in the cathode from the calcium cell appears continuous and "glassy"

nd would be expected to interfere with transport and conduction processes

in the electrolyte phase. Studies are currently in progress to evaluateand overcome this problem.

STAMP

Downgrad ing/Declassi fi-

cat ion

In format ionon first STAMP Security Classification

at Oherepaper l' 4--

-Ipa

ARMXY SC Il:NC17 CONVI KNCE -PAP1r V.r--, -S- :;TAMP Security

n svyt last F ss - l fssi ficat i oae of lu- here

_or(s) w GILMAN, WADE, BINDER

tart her(four al l[pa:., s --'1 -

f i r"

: 1;fI t i l. . ..

Fis iea. Li thium anode b. Calcium anode!

OnFirst .- ineIi '

--gFijure , e 1 ct-ro -m Icog--isst2Ci (3260-it magn-ification).

Finally it may be noted that on thermodynamic grounds, approximiately500 millivolts of terminal voltage can be "recaptured" tf the calcium

electrode can be made to operate more reversibly. The present perform-!

rnte, sli 1;l r to that observed in the lower-potential Ca/SOCI 2 sy -:r!Z1

(8,9) can be attributedl to "film pol.arizaton" in the protective coating

(probably CaCI12 , according to the react-ion in Table 1) forred by li1mitedspontaneous reaction of thde with the solvent. Improved terminal

voltage can be expected if it is discovered how to miiize the latter

coating nbile still retaining its protective properties.

REFERENCES

1. S. Gilman and W. Wade, Jr., Elcctrochem Soc., 127, 1427 (1980).2. W. K. Behl, J. A. Christopulos, M. Ramirez and S. Gilman, J. Electro-

chem Soc., 120, 1619 (1973).3. F. Marakar, "High Rate Lithium-Sulfuryl Chloride Battery Technology,"

First Quarterly Re~Iort, Contract DAkK20-81-C-0420 (ERADCOM), CouldLabs, March 1982.

4. J. J. Auborn and N. Marincic, "Power Sources 5," D. H. Collins, Editorp. 683, Academic Press, London (1975).

5. J. J. Auborn, R. D. Bezman, K. W. French, A. Heller and S. F. Liebermanin Proceedings of 26th Power Sources Symposium, p. 45 (1974).

6. "Encyclopedia of Chemical Technology," Vol. 14, K. Othmen, Editor,p. 45 (1969).

7. W. K. Behl, J. Electrochem Soc., 127, 1444 (1980).8. R. L. Higgins, Proceedings of the 29th Power Sources Conference, p. 14

(1980).SANP 9. K. J. Staniewicz, J. Electrochem Soc., 127, 782 (1980).l)owngrad i i_/_L\c lassili-cationInformationon first STAMP Security Classification

.-_,.-of herepaper nl - --

page 1)